Dean Roemmich

Photo credit: Scripps Insti-

tution of Oceanography

Lee-Lueng Fu

Lee-Lueng Fu and Dean H. Roemmich

Sea level rise is an indicator of the extent of the warming of the Earth’s climate as well as a major threat to the world’s coastal zones. The rate of the rise of the global mean sea level has been accelerating since the Industrial Revolution, reaching over 3 mm/yr at present. New technologies developed over the past 25 years have enabled great strides in monitoring global sea level with both spaceborne and in situ sensors, revealing information that is useful for understanding the phenomenon and predicting its evolution.

Introduction

Over the geological history of Earth, sea level has varied by hundreds of meters as a result of tectonic and climatic processes. The current ice age cycle started about 3 million years ago, with a 100,000-year cycle in the past 1 million years primarily caused by fluctuation in the Earth’s orbit around the Sun. During the most recent glacial maximum 25,000 years ago, sea level was about 130 meters below the present level. During the deglaciation that began 20,000 years ago, sea level began rising rapidly at 1 cm/year until 7,000 years ago, when the rate stabilized to 0.2 mm/yr (Carlson and Clark

Monitoring Global Sea Level Change from Spaceborne and In Situ Observing Systems

Recent advances in the technology for observing the

ocean from spaceborne and in situ sensors make it

possible to monitor the rise of the global sea level with

unprecedented accuracy.

Lee-Lueng Fu (NAE) is Ocean Surface Topography Mission project scientist at Jet Propulsion Laboratory, California Institute of Technology, Pasadena. Dean Roemmich (NAE) is distinguished professor of oceanography at Scripps Institution of Oceanography, La Jolla, CA.

55FALL 2018

2012). In the 20th century, however, the rate rose ten-fold to 2 mm/yr and in the past 20 years it accelerated to 3 mm/yr, a third of the rate during the maximum deglaciation (Church and White 2011).

Causes and ImpactsThe recent rise of the global sea level is caused by thermal expansion from the warming of the ocean and the melt-ing of ice on land. During the past 20 years the former has accounted for about one third of the rise and the latter about two thirds according to the 5th Assessment Report of the Intergovernmental Panel on Climate Change (IPCC; Church et al. 2013). Based on the high end of the projected warming of the planet in the coming decades (compared with the period of 1986–2005), the IPCC estimated the rise of the global mean sea level (GMSL) at 52–98 cm by 2100, with a rate of 8–16 mm/year during 2081–2100.

The rising GMSL has already significantly affected the 10 percent of the world’s population living at eleva-tions lower than 10 m near the ocean. The threats to low-lying islands of the Maldives and the coastal zones of Bangladesh are well known. In New York City the last 7 percent of the storm surge from Hurricane Sandy affected 11.4 percent more people and 11.6 percent more housing units, and caused 24 percent more total property damage, than it would have without the sea level rise of the past 100 years (Leifert 2015). The high-er sea level makes many coastal cities prone to flooding during high tides and increases the frequency of so-called nuisance floods (e.g., Kruel 2016).

Evolution of Efforts to Measure Sea Level RiseBefore the advent of satellite remote sensing and its global coverage, it was not straightforward to measure the GMSL using only tide gauges, whose sparse and uneven coverage resulted in unknown sampling errors in efforts to determine the GMSL. In the late 1960s the concept of using a radar altimeter on an Earth-orbiting satellite to measure sea surface height was developed, and the first such altimeter, launched in the 1970s, demonstrated space observations of sea level. Within two decades the accuracy and precision of satellite altimetry were sufficient to determine the GMSL and small changes in it.

Two other advances important to understanding changes in the GMSL emerged in the 1990s. One was the deployment of autonomous profiling floats in the ocean to measure the temperature and salinity of

the water column globally through an international program called Argo.1 The technique enables determi-nation of water density as a function of depth and its effects on changes in sea level. For example, rising tem-perature would raise sea level via thermal expansion.

The other development was the launch of satellites to measure Earth’s changing field of gravity resulting from the changing distribution of mass near the planet’s surface. This satellite mission, called GRACE (Gravity Recovery and Climate Experiment2), has determined the contribution of changes in the mass of the water column to sea level change.

Together, satellite altimetry, Argo, and GRACE pro-vide an observing system for determining changes in the global sea level and their causes: changes in water density and mass. The resulting information has revo-lutionized both the capability to monitor small signs of global sea level rise and understanding of the physical processes needed to project future changes.

Satellite Altimetric Measurement of Sea Level Change

The measurement configuration of satellite altimetry is illustrated in figure 1. A radar altimeter on an orbit-ing satellite, operating at microwave frequencies of 10–35 GHz, sends short pulses to the sea surface and receives the return signals, and the round-trip travel time is used to determine the distance between the satel-lite and sea surface. With the height of the satellite rela-tive to Earth’s center of mass (~1,000 km) determined by the technology of precision orbit determination, it is possible to calculate the geocentric sea level (the height of sea surface relative to Earth’s center of mass).

1 https://sealevel.nasa.gov/missions/argo2 www2.csr.utexas.edu/grace

Hurricane Sandy caused 24 percent more total property damage than

it would have without the sea level rise of the past 100 years.

TheBRIDGE56

The shape of the sea surface is dictated to a large extent by the variation of Earth’s gravity field at its sur-face, caused by the uneven structure and density of the lithosphere, the upper layer of solid earth. The relief of the sea surface is within a couple of meters of a surface of constant gravity, called the geoid, whose relief relative to the reference ellipsoid has a range of about 200 m. The deviation of the sea surface from the geoid is called the ocean dynamic topography.3

Early EffortsThe first spaceborne radar altimeter was aboard the Sky-lab missions in the early 1970s (Krishen 1975), followed by the series of the Geodetic Earth Orbiting Satellite (GEOS) missions (Stanley 1979) in the mid-1970s to refine the measurement to an accuracy necessary to determine the shape of the geoid to a few meters. Not until the launch of Seasat in 1978 (Born et al. 1979) was the accuracy of satellite altimetry sufficient—within a few centimeters—for measuring the variability of ocean

3 The horizontal gradient of the dynamic topography is propor-tional to the part of ocean surface circulation that is balanced by the Earth’s rotation (the Coriolis force). This flow compo-nent, the geostrophic circulation, is responsible for the large-scale transport of ocean mass and heat, providing a fundamental moti-vation for the development of satellite altimetry in addition to the determination of sea level.

surface topography; the uncertainty of the satellite’s radial height was about 1 m over scales of 10,000 km.

Motivated by the desire to determine the ocean general circulation at the scales of the ocean basin, a satellite mission called TOPEX/Poseidon (T/P) was developed jointly by NASA and the French space agen-cy, CNES (Fu et al. 1994). Launched in 1992, T/P was able to measure the ocean dynamic topography and geo-centric sea level to centimetric accuracy, representing a remarkable achievement in improving the capability of satellite altimetry by a factor of 100 in 25 years.4

Increased AccuracyA series of satellite missions carrying radar altimeters has been on orbit since the 1990s, yielding a continuous record of global sea level and ocean dynamic topogra-phy. Although the early mission development was not focused on the GMSL, whose required accuracy was too daunting a task in the 1980s, the breakthrough enabled by T/P has made the GMSL a key objective of current altimetry missions.

A time series shows the change of the GMSL from 1993 to 2018 (figure 2; Nerem et al. 2018). The record is the result of extensive calibration and validation on four missions (T/P and its follow-on Jason series). The GMSL during this period is estimated at 3.1 ± 0.4 mm/yr (the uncertainty is largely from calibration against the global tide gauge network). Other studies have placed the uncertainty at 0.3–0.5 mm/yr (Ablain et al. 2017). Before satellite measurement, it was not possible to rig-orously estimate the GMSL and its uncertainty.

How good is the current capability of determining the rate of GMSL rise? The rate of acceleration to reach 70 cm rise above today’s GMSL by 2100 (roughly the middle of the IPCC’s high-end projection) is ~0.1 mm/yr2. The recent study by Nerem and colleagues (2018) concluded, with high statistical confidence, that accel-eration of this magnitude—0.08 mm/yr2—is already observed in the present 25-year record of altimetric GMSL. This finding suggests that the current capability of monitoring the GMSL meets society’s needs for rigor-ous assessment of future threats of global sea level rise.

Argo Float Measurement of Heat-Induced Sea Level Change

The profiling float (figure 3) is a free-drifting and wholly autonomous ocean instrument developed by Davis

4 The tremendous progress of satellite altimetry is described in Fu and Cazenave (2001) and Stammer and Cazenave (2018).

FIGURE 1

FIGURE 2

FIGURE 1 Measurement configuration of satellite altimetry. Reprinted with permission from Fu (2001).

57FALL 2018

and colleagues (2001) for research use during the 1990s World Ocean Cir-culation Experiment. The instruments adjust their buoyancy and hence their depth by pumping mineral oil between a reservoir in the float’s pressure case and an external bladder. Changes in the volume of a float by a few percent cause buoyancy variation sufficient to keep the float on the sea surface, make it neutrally buoyant at inter-mediate depths, or send it to the ocean bottom. A sen-sor package mounted on top of the float collects profile measurements of tempera-ture, salinity, and pressure. The float cycles between the sea surface and a depth of 2,000 m every 10 days, transmitting its GPS posi-tion and profile data via Iridium satellites when it surfaces.

Before the development of profiling floats, subsur-face ocean temperature data could be collected only when a ship was present or from fixed-point moorings. Resulting datasets were very sparse and irregular in space and time, with most data obtained in the Northern Hemisphere, near continents, and in summer. The profiling float was a revolutionary instrument for oceanography because it provides high-quality data anywhere, any time.

In 1997 an international group of scientists proposed the installation of a global array, named Argo, consisting of 3,300 profiling floats (Argo Science Team 1998). A

primary motivation was to assess climate variability and change, including GMSL rise due to ocean warming. The first Argo floats were deployed in 1999; by 2007 there were over 3,000 distributed globally, and about

FIGURE 1

FIGURE 2

FIGURE 2 Time series of the (global) mean sea level (MSL) in mm from the TOPEX/Poseidon mission and its follow-ons: Jason-1, Jason-2, and Jason-3, 1993–2018. From University of Colorado (http://sealevel.colorado.edu).

10

FIGURE 3

FIGURE 3 A profiling float is deployed by RV Southern Surveyor. Photo by Alicia Navidad, CSIRO. Credit: Argo Program.

TheBRIDGE58

3,800 have been maintained for the past decade. Begun in 2006 as a multinational effort, Argo is sustained by national programs in more than 25 countries (figure 4).

Recent advances in profiling float and sensor tech-nologies now make it possible to sample to the ocean bottom at depths of up to 6,000 m (Deep Argo; Zilber-man 2017) and to sample additional parameters such as dissolved oxygen, nitrate, pH, and biooptical properties (Biogeochemical Argo; Johnson et al. 2009). The global Argo array greatly increases the accuracy of the earlier estimates of thermosteric GMSL rise based on the sparse pre-Argo datasets.

The thermal expansion coefficient for seawater varies with temperature and pressure. The top 1,000 m of sea-water at the equator, if warmed uniformly by 0.1°C, would expand in height by 1.8 cm, while at 60°S the expansion would be 0.8 cm. Sea-Bird electronic sensors on Argo floats measure temperature, salinity, and pres-sure with high accuracy (.002°C, .01 psu, and 0.1 per-cent respectively). Given the sensor accuracy and the

fact that float-to-float differences are mostly random, errors in large-scale temperature variability are mainly due to the array’s spatial coverage (figure 4). That is, limited data coverage in the deep ocean and spatial inhomogeneity are the main source of error in ther-mosteric GMSL estimates. Deep ocean warming below Argo’s present depth limit of 2,000 m is estimated to account for 0.1 mm/yr to GMSL rise in 1993–2010 (Purkey and Johnson 2010), or about 10 percent of the 0–2,000 m GMSL thermosteric contribution, which was 1.0 ± 0.2 mm/yr in 2004–16 (Thompson et al. 2017).

The Argo program has produced over 12 years of global data, which are freely available via the internet (www.argo.net) in near real time and as research qual-ity after 1 year. A monthly interpolated version (Roem-mich and Gilson 2009) is used to estimate the trend in steric height of the sea surface relative to 2,000 decibars as a function of location (figure 5). The same calcula-tion is made over the same time interval with satellite altimetric sea level data.

11

FIGURE 4

FIGURE 4 Location of 3,825 operational Argo floats, May 2018. Colors represent the 25 participating national Argo programs plus Euro-Argo. Source: Argo Information Center, www.jcommops.org.

59FALL 2018

Global patterns of the trend in altimetric sea sur-face height and steric height are compared in figure 5. Two aspects of this comparison are evident. First, the global mean of the altimetric height trend (upper panel, 3.3 mm/yr) is greater than that of steric height (lower panel, 1.3 mm/yr), and this 2 mm/yr difference is attrib-utable to mass (discussed below). Second, discounting the difference in global means, the pattern of regional variability in the altimetric and steric height trends is similar, indicating that the mass trend, while large, is more spatially uniform than the steric height trend.

Regional trends in steric height (figure 5) can be

much larger than the global mean.5 These massive redis-tributions of warm ocean water are largely wind driven and can increase the rate of sea level rise regionally for years or longer; some regions of large steric sea level rise, such as along 40°S, are known to have multidecadal timescales (Roemmich et al. 2016). In other cases the warm anomalies represent interannual changes that appear trend-like in the 12-year time series; for exam-

5 Changes in global mean steric height due to salinity are too small to detect in the 12-year time series, but there are regional examples, particularly at high latitude where the impact of salinity on steric height is significant.

12

FIGURE 5

FIGURE 5 Trends in altimetric sea surface height (top panel, mm/year) and 0–2,000 m Argo steric height (bottom panel, mm/year), for the period 2006–18. Sea surface height trends are based on the multimission dataset of Scharroo et al. (2013). Steric height trends are based on the gridded Argo dataset described by Roemmich and Gilson (2009).

TheBRIDGE60

ple, the large maximum in the eastern equatorial Pacific is attributable to a major El Niño episode in 2015–16. A longer Argo time series is needed for effective separa-tion of interannual and decadal variability.

GRACE Measurement of Mass-Induced Sea Level Change

The original concept of determining the ocean circula-tion from space required not only satellite altimetry but also the measurement of the geoid for computing the ocean dynamic topography. After many variations of the design of a spaceborne gravity mission, the concept of GRACE emerged to measure the minute change of gravity experienced by two spacecraft on orbit track-ing each other using a microwave link (NRC 1997). GRACE can determine not only the near surface grav-ity field of Earth but also its change with time.

How GRACE WorksTo be sensitive to the spatial variability of Earth’s gravi-ty associated with the structure and density of the upper layers of the solid earth, the two GRACE spacecraft, launched in 2002, were placed in a near-Earth orbit of ~500 km, 220 km apart (figure 6). GRACE measures

a “biased range” between the two spacecraft by track-ing the carrier phase of a K and Ka band microwave signal (Tapley et al. 2004). The bias is constant over long periods so that the range change is very accurately measured. The wavelength of the K/Ka signals is about 1 cm. Careful design and averaging over several seconds yield accuracy between 1 part per thousand and 1 per ten thousand of the wavelength, corresponding to a few microns in range every 5 seconds. The data are often processed as range rate, which is good to about 0.1 µm/sec for 5-second averages. Since the measurements are one way from each spacecraft, it is important to know the time of transmission and reception on each space-craft and to be able to synchronize those to about 100 picoseconds. The accuracy of GRACE in measuring the change of gravity can be expressed as that associated with the mass of water of 1–2 cm thickness over a circle with a radius of 300 km (Landerer and Swenson 2012).

Uses of GRACE DataThe gravity measurement of GRACE enables determi-nation of the change of mass over large ocean areas and, when integrated globally, GRACE data reveal the con-tributions of ice melt in the change of the GMSL. As

13

FIGURE 6

FIGURE 6 Schematic illustration of twin GRACE spacecraft tracking each other in orbit. The exaggerated surface features of Earth are intended to reflect the planet’s varying gravity field. CSR = Center for Space Research; DLR = Deutsches Zentrum für Luft- und Raumfahrt e.V.; GFZ = GeoForschungsZentrum; GRACE = Gravity Recovery and Climate Experiment; NASA = National Aeronautics and Space Administration. Image credit: GRACE Mission Team.

61FALL 2018

shown in figure 7, the com-bination of the steric sea level change (caused by the change of ocean density), estimated from the Argo data, with the barystatic sea level change (caused by the change of ocean mass) matches the altimetry mea-surement of total sea level change fairly well. The synergy of the three mea-surement systems creates an opportunity for cross-validation of the difficult and important measure-ment of the GMSL, which is an indicator of both the extent of climate change and impacts on humans.

GRACE data have revealed the rate of the melting of ice sheets on Greenland and Antarctica (Velicogna 2009), whose potential massive breakup and melting are the pri-mary sources of the looming threat of sea level rise. The fragile West Antarctic Ice Sheet holds enough water to raise the GMSL by 5 m, in contrast to the 0.5 m capacity of the warming of the ocean. Distribution of the melt water will be uneven geographically, because of changes both in Earth’s gravity due to the massive ice loss in the polar regions and in the ocean’s density and circulation due to climate change. The greatest magnitude is projected to account for 20 percent of the change of the GMSL (Slangen et al. 2012). Such vari-ability in regional sea level change is of great concern in efforts to plan for coastal adaptation. Figure 8 dis-plays the 2005–14 pattern of sea level change due to the melting of polar ice sheets estimated from the GRACE data (Adhikari and Ivins 2016), qualitatively similar to the projection of Slangen and colleagues, given the relatively short data record.

An interesting feature is that the reduced grav-ity caused by the mass loss of the polar ice sheets has caused sea level to fall in regions close to the source of the melting in Greenland and Antarctica and rise

elsewhere. As opposed to the interannual pattern of sea level change shown in figure 5, the pattern of change from ice melting has a much longer time scale, which is most relevant to coastal planning and decision making.

Concluding Remarks

Advances in technology for observing the ocean from spaceborne and in situ sensors make it possible to moni-tor the rise of the global sea level with unprecedented accuracy.

Signs of accelerating GMSL rise over the past two decades have been detected by the satellite radar altim-etry system with high statistical confidence. The prior-ity of the international space community is to maintain the present capability to monitor current and future changes. Spaceborne gravity measurement allows detec-tion of changes in GMSL from water mass exchange between the ocean and the rest of Earth, primarily from ice melting. The in situ float system detects changes in water temperature and salinity, and hence density, with

14

FIGURE 7

FIGURE 7 Altimetry-based estimate of the global mean sea level (GMSL; black); Argo-based steric sea level (green); GRACE-based ocean mass (in equivalent of sea level, blue), January 2005–December 2014. The red trendline represents the sum of the steric and ocean mass components. An arbitrary vertical offset was applied to the green and blue lines for clarity. CCI = European Space Agency Climate Change Initiative; GRACE = Gravity Recovery and Climate Experiment. Reprinted by permission from Springer (Ablain et al. 2017).

TheBRIDGE62

the long-term trend primarily due to the warming of the climate.

The various systems together not only provide infor-mation to understand the causes of sea level rise but also enable cross-checking for consistency to ensure the accuracy of the observations, which are critical for deal-ing with climate change.

References

Ablain M, Legeais JF, Prandi P, Marcos M, Fenoglio-Marc L, Dieng HB, Benveniste J, Cazenave A. 2017. Satellite altimetry-based sea level at global and regional scales. Sur-veys in Geophysics 38(1):7–31.

Adhikari S, Ivins ER. 2016. Climate-driven polar motion: 2003–2015. Science Advances 2(4):e1501693.

Argo Science Team. 1998. On the Design and Implementa-tion of Argo, a Global Array of Profiling Floats. Report 21. Qingdao: International CLIVAR Global Project Office.

Born GH, Dunne JA, Lame DB. 1979. SEASAT mission over-view. Science 204:1405–1406.

Carlson AE, Clark PU. 2012. Ice sheet sources of sea level rise and freshwater discharge during the last deglaciation. Reviews of Geophysics 50(4):RG4007.

Church JA, White NJ. 2011. Sea-level rise from the late 19th to the early 21st century. Surveys in Geophysics 32(4–5):585–602.

Church JA, Clark PU, Cazenave A, Gregory JM, Jevrejeva S, Levermann A, Merrifield MA, Milne GA, Nerem RS,

Nunn PD, and 4 others. 2013. Sea level change. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, eds Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM. Cambridge and New York: Cambridge University Press.

Davis RE, Sherman JT, Dufour J. 2001. Profiling ALACEs and other advances in autonomous subsurface floats. Journal of Atmospheric and Oceanic Technology 18:982–993.

Fu L-L. 2001. Ocean circulation and variability from satellite altimetry. In: Ocean Circulation and Climate, eds Siedler G, Church J, Gould J. San Diego: Academic Press. pp 141–172.

Fu L-L, Christensen EJ, Yamarone CA, Lefebvre M, Menard Y, Dorrer M, Escudier P. 1994. TOPEX/Poseidon mission overview. Journal of Geophysical Research 99:24369–24381.

Johnson KS, Berelson WM, Boss ES, Chase Z, Claustre H, Emerson SR, Gruber N, Körtzinger A, Perry MJ, Riser SC. 2009. Observing biogeochemical cycles at global scales with profiling floats and gliders: Prospects for a global array. Oceanography 22(3):216–225.

Krishen K. 1975. The significance of the S-193 Skylab experi-ment using preliminary data evaluation. NASA Contract Report, NASA-CR-150989. Houston: Lockheed Elec-tronics Company.

Kruel S. 2016. The impacts of sea-level rise on tidal flooding in Boston, Massachusetts. Coastal Research 32(6):1302–1309.

15

FIGURE 8

FIGURE 8 Sea level change due to climate-induced mass redistribution from polar ice melt, impact on the oceans. Estimated from the Gravity Recovery and Climate Experiment (GRACE), April 2002–March 2015. Reprinted under CC-BY-NC open access license from Adhikari and Ivins (2016).

63FALL 2018

Landerer FW, Swenson SC. 2012. Accuracy of scaled GRACE terrestrial water storage estimates. Water Resources Research 48(4):W04531.

Leifert H. 2015. Sea level rise added $2 billion to Sandy’s toll in New York City. Eos 96, March 16.

Nerem RS, Beckley BD, Fasullo JT, Hamlington B, Masters D, Mitchum GT. 2018. Climate-change–driven accelerated sea-level rise detected in the altimeter era. Proceedings of the National Academy of Sciences, February 12.

NRC [National Research Council]. 1997. Satellite Gravity and the Geosphere: Contributions to the Study of the Solid Earth and Its Fluid Envelopes. Washington: National Academy Press.

Purkey SG, Johnson GC. 2010. Warming of global abyssal and deep Southern Ocean waters between the 1990s and 2000s: Contributions to global heat and sea level rise bud-gets. Journal of Climate 23(23):6336–6351.

Roemmich D, Gilson J. 2009. The 2004–2008 mean and annual cycle of temperature, salinity, and steric height in the global ocean from the Argo program. Progress in Oceanography 82(2):81–100.

Roemmich D, Gilson J, Sutton P, Zilberman N. 2016. Multi-decadal change of the South Pacific gyre circulation. Jour-nal of Physical Oceanography 46(6):1871–1883.

Scharroo R, Leuliette E, Lillibridge J, Byrne D, Naeije M, Mitchum G. 2013. RADS: Consistent multi-mission

products. Proceedings of the Symposium on 20 Years of Progress in Radar Altimetry (ESA SP-710), September 20–28, 2012, Venice.

Slangen ABA, Katsman CA, van de Wal RSW, Vermeersen LLA, Riva REM. 2012. Towards regional projections of twenty-first century sea-level change based on IPCC SRES scenarios. Climate Dynamics 38(5-6):1191–1209.

Stammer D, Cazenave A. 2018. Satellite Altimetry over Oceans and Land Surfaces. Boca Raton: CRC Press.

Stanley HR. 1979. The Geos 3 Project. Journal of Geophysi-cal Research 84(B8):3779–3783.

Tapley BD, Bettadpur S, Watkins M, Reigber C. 2004. The gravity recovery and climate experiment: Mission over-view and early results. Geophysical Research Letters 31(9):L09607.

Thompson PR, Merrifield M, Leuliette E, Sweet W, Chambers D, Hamlington BD, Jevrejeva S, Marra JJ, Mitchum G, Nerem RS. 2017. Sea level variability and change. Bulletin of the American Meteorological Society: State of the Climate in 2016 8(98):79–81.

Velicogna I. 2009. Increasing rates of ice mass loss from the Greenland and Antarctic ice sheets revealed by GRACE. Geophysical Research Letters 36(19):L19503.

Zilberman NV. 2017. Deep Argo: Sampling the total ocean volume. Bulletin of the American Meteorological Society: State of the Climate in 2016 8(98):73–74.